Inspiration

"Aren't you Chinese? You must be good at math."

I grew up hearing this stereotype, yet I failed math constantly and internalized the belief that I simply wasn't "wired" for STEM. That fear dictated my life choices—until GRE prep. In two months, my tutor (now my husband) didn't teach me new formulas; he rebuilt my confidence. I scored in the global top percentile.

That was my epiphany: every child deserves someone who believes in them more than they believe in themselves.

After 12 years as an EdTech founder serving 2,000+ families and earning my PhD in Educational Psychology, I saw the same pattern: bright girls shutting down before they even try. Math anxiety literally consumes working memory (Ramirez et al., 2013). I built Math Muse to be the digital big sister I wish I had—a social-emotional intelligence (SEI) math-learning tool that rewrites failure narratives from "Math is just not for me" to "I can try another method."

What it does

Math Muse is a story-driven, Gemini 3-powered platform that reduces math anxiety and rebuilds confidence specifically for girls.

• Narrative Adventures: Users become Vera Wang designing a royal wedding dress or an environmental scientist saving coral reefs—solving real geometry and ratio problems to advance the plot.
• Gemini 3 as Empathetic Mentor: Instead of red Xs, Gemini reframes errors:
  > "That was a calculation slip, not a logic error. Let's try another method together ✨"
  It uses 1M-token context to remember emotional history across sessions.
• Anxiety-Adaptive Engine: Monitors hesitation time, hint usage, and sentiment to keep users in Vygotsky’s Zone of Proximal Development + Csikszentmihalyi’s flow state.
• Growth-Mindset Reframing: Every mistake is attributed to controllable factors (strategy, effort) rather than fixed ability—directly countering stereotype threat.

How we built it

As a non-traditional coder with a PhD in Educational Psychology, I used Gemini 3 as my co-engineer:

• Gemini 3 Pro + Live API → Core reasoning engine for real-time empathetic responses, long-context emotional memory, and multimodal input (text + future voice/image).
• Expectancy-Value Theory Implementation
  Motivation = Expectancy × Value
  Gemini dynamically adjusts narrative relevance (Value) and success scaffolding (Expectancy) via system prompts grounded in decades of motivation research.
• Verification Layer → Forces Gemini to solve math symbolically first (using code execution) before narrating, eliminating hallucinations on arithmetic.
• Frontend → React + Framer Motion for animated "confidence garden" progress visualization.
• Deployment → Google AI Studio → instant public prototype; Vercel for production.

Challenges we ran into

• Hallucinated Math: Early versions confidently gave wrong answers wrapped in beautiful stories. Solved with mandatory code-execution verification.
• Balancing Story vs. Rigor: Too much narrative distracted from learning. Iterated until math became the tool to advance the beloved character’s goal.
• Solo "Vibe Coding": Translating psychological frameworks into prompt engineering required hundreds of iterations—but proved domain experts can build with AI.

Accomplishments that we're proud of

• Solo-built a Gemini 3-native SEI agent that achieves what traditional tutors struggle with: real-time emotional regulation at scale.
• First known implementation of Expectancy-Value Theory as live system prompts in a consumer math app.
• Migrated from no-code (Base44) to Gemini 3-native in under 72 hours—proving the power of the new API.

What we learned

Gemini 3 isn’t just a reasoning engine—it can be an empathy engine when guided by psychological science. With the right system instructions, it becomes the patient, believing mentor every anxious learner needs.

What's next for Math Muse

• Feb–Mar 2026: 50-girl pilot with pre/post math anxiety + growth-mindset surveys
• Apr 2026: Fine-tune Gemini on "Math Therapy" interaction data
• 2026 Goal: 10,000 users and peer-reviewed evidence that SEI-powered learning closes the STEM gender confidence gap

Math Muse isn’t just teaching math—it’s healing the relationship girls have with failure.

NittanyAI #Gemini3 #EducationEquity #GrowthMindset

Built With

• ai-&-machine-learning:-google-gemini-api-core-ai-engine-powering-interactive-math-tutoring
• personalized-problem-generation